Electronic devices generate significant amounts of thermal energy during operation. The functional lifetime of electronic devices is significantly diminished by excess heat buildup. Therefore, a number of methods have been presented to remove thermal energy from electronic devices and reject it into an external environment. Since the beginnings of electronic devices, air movement over these devices has been the primary means of heat removal. For example, in the early large-scale computing systems of the 1940s and 1950s, heat dissipation consisted primarily of ventilation apertures in housings, followed by ambient-air fans and blowers which cooled by forced air convection. Even today, refined versions of these early air-based heat dissipation systems are the most common means of electronic device and computer systems cooling. In air-based heat dissipation systems, air within a device enclosure is heated by the electronic device and internal fans expel heated air into the immediate environment around the device. The environment around the device is typically maintained with regards to temperature, humidity, and particulate matter, by using compression-based heat exchange with the outside environment. This process is effective and in common use for non-stop electronic devices such as computer servers. Although this process is effective, it is complex process with a number of systems that must be constantly maintained to produce the desired environment thus having high construction and operational costs. For example, air-based cooling relies on a) the proper operation of fans to circulate air inside the device enclosure, in the server room, and in outside condensers, b) a very clean environment free of most dust and particulates, c) proper humidity control, and d) costly “white space” in the server room to allow human access to electronic devices for repair and maintenance. Air based cooling faces significant risks from a) internal fan and cooling failures, b) server room cooling failures and inconsistencies, c) fire control systems, d) unauthorized human access, e) maintenance failures and mistakes, and f) natural disasters. Taken together, these factors typically require specialized and costly installation space for electronic devices such as computer servers. Further, air-based cooling of electronic systems can double the total amount of electrical energy required to operate these systems, resulting in a costly and wasteful means of operating such systems.
Noting the inefficiencies and problems with air-based heat dissipation, designs begin to arise in the 1960s and 1970s that took advantage of the much higher thermal conductivity of liquids, which typically conduct heat ten to one hundred times more rapidly than gases. Liquid vapor cooling of individual semiconductors and other solid state components was disclosed by Davis in U.S. Pat. No. 3,270,250, and in U.S. Pat. No. 3,524,497, Chu et. al. disclose a double-walled container for component-level electronics, with liquid flow in the space between the walls. The predominance of such designs focused on component level cooling of larger systems.
As individual CPU processing speed and power increased during the 1980s, inventors continued to disclose methods for additional cooling capability in electronic assemblies. Many of these disclosures related to component level cooling, but a few began to focus on system level liquid cooling. Cray, in U.S. Pat. No. 4,590,538 (1986), discloses a means of immersing an entire electronic assembly in coolant liquid, and circulating the liquid out of the assembly container for the purpose of thermal energy removal. Numerous other methods of liquid cooling of components and component assemblies continued to be disclosed throughout the 1990s. In the late 2000s, the liquid cooling designs from the 1980s and 1990s were applied to individual servers and computing systems. These innovations were followed by modifications and improvements which incorporated liquid cooling elements into the structural design of computing systems rather than individual modules or computing units. For example, in U.S. Pat. No. 8,351,206, Campbell et. al. disclose a liquid-cooled electronics rack with immersion-cooled electronics and a vertically mounted vapor condensation unit attached to or adjacent to the electronics rack.
Olsen, et. al. describe in U.S. Pat. No. 8,416,572 a design for multiple electronic devices connected in an array, thermally coupled to a flowing liquid. In U.S. Pat. No. 8,467,189 and related following patents Attlesey discloses designs for an array of rack-mounted plurality of cases for electronics systems; each case contains a dielectric fluid for heat conduction, and the rack system incorporates a manifold for liquid circulation through the plurality of cases, with a pump and heat exchanger incorporated into the fluid circulation loop. Best et. al. disclose, in U.S. Patent Application 2011/0132579 a design in which a series of horizontally oriented computer server racks are submerged in a liquid tank containing a dielectric cooling fluid that is circulated from the tank to a remote heat exchanger and back into the tank.
One of the significant improvements of liquid cooling over air cooling is the ability to transport heat from the electronic device or system directly to the heat rejection environment without significantly affecting the human inhabited space in the server room thus dramatically increasing the heat transport efficiency while reducing the number of cooling processes and preventing excess heat diffusion. However, these processes have not seen widespread adoption for one or more possible reasons. Component level liquid cooling designs tend to introduce significant complexity to operations and maintenance while increasing server room risks to coolant leaks and failures. System level liquid cooling designs reduce the overall number of cooling interconnects, but have similar problems. To further complicate the liquid cooling server room installations, liquid cooled systems require new server room procedures, operations, and training and expose owner and operators to additional liabilities from liquid damage. And notably, production electronic devices and servers are rarely available in liquid cooling configurations. Succinctly, the cost savings associated with current liquid cooling designs are overshadowed by the increased costs of purchasing, constructing, and operating liquid cooled servers and solutions.
Significantly, it is the widespread usage of virtualized computing resources that is allowing greater innovation and deployment of fluid cooled electronic devices and servers. Virtualization of data resources allows data to be stored on many redundant devices. Virtualization of compute resources allows the functional compute unit of a “server” to become a software unit that can be moved from one physical computer to another. Individual electronic devices and servers may fail over time, but the virtualized nature of software based compute and storage units mean that an individual failures only slightly decreases the overall capability of a collection of servers but in no way compromises the data processing, storage, and communication functions as a whole. Therefore, since it is no longer necessary to maintain or repair a specific physical server in order to maintain a given operation, fluid cooling of electronic devices in a sealed enclosure is enabling cost reductions, operational efficiencies, increased security, and extended longevity of electronic devices and servers.
The innovations as disclosed herein overcome problems inherent to both traditional air-cooled and liquid-cooled electronic devices and systems. Significant benefits comprise a) high efficiency cooling and heat exchange reducing overall energy usage by up to 50%, b) no maintenance required, c) devices and systems can be installed in almost any environment such as a traditional data center, high rise office, industrial building, offshore installation, underground installation, and ambient air data center, d) increasing server density up to 3× the current high-density server deployments thus reducing the amount of server room space required, e) improved physical security, f) improved EMI/RFI security, g) decreased labor costs, h) more protection against disasters such as fire, hurricane, and earthquake, i) fewer maintenance failures and mistakes, j) tamper-resistant to unauthorized human access, k) reduced or eliminated damage due to fire control systems, l) nearly silent in operation, m) internal components have cooler average temperature that will increase the life of the system, and n) impervious to environmental factors such as dust and humidity.
These and other benefits disclosed herein combine together to create entirely new classes of solutions. For example, innovation in the fluid cooling of electronic devices as disclosed herein, and innovations that allow for a broader range of installation environments are disclosed by Smith in U.S. Patent Appl. No. 2015/0000319 (January 2015) are challenging the assumptions and designs of data centers and server rooms.
Unless specifically stated as such, the preceding is not admitted to be prior art and no statement appearing in this section should be interpreted as a disclaimer of any features or improvements listed.